fix(node): release a deleted node's metric series and HTTP client

Deleting a node must free what the master keeps per node in memory.

Delete dropped the node's cpu and mem series but not netUp and netDown,
which the heartbeat records too, so each deleted node leaked two tiered
histories. It now drops every NodeMetricKeys entry.

InvalidateNode, called on node edit, disable and delete, cleared only the
cached Remote. The pooled HTTP client and its transport stayed cached until
a later call for the same node pruned them, which a deleted node never
makes. InvalidateNode now drops those too, outside the manager lock; an
edited node pays one fresh handshake on its next call.

internal/web/service/node_delete_cleanup_test.go

@@ -0,0 +1,43 @@
+package service
+
+import (
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/mhsanaei/3x-ui/v3/internal/database"
+	"github.com/mhsanaei/3x-ui/v3/internal/database/model"
+)
+
+// Deleting a node dropped only its cpu and mem series, while the heartbeat also
+// records netUp and netDown, so each deleted node leaked two histories for good.
+func TestDeleteNodeDropsEveryMetricSeries(t *testing.T) {
+	setupConflictDB(t)
+	node := &model.Node{Id: 9101, Name: "gone", Address: "127.0.0.1", Port: 2096, ApiToken: "tok", Enable: true, Status: "online"}
+	if err := database.GetDB().Create(node).Error; err != nil {
+		t.Fatalf("create node: %v", err)
+	}
+	ns := NodeService{}
+	if err := ns.UpdateHeartbeat(node.Id, HeartbeatPatch{
+		Status: "online", LastHeartbeat: time.Now().Unix(), CpuPct: 1, MemPct: 2, NetUp: 3, NetDown: 4,
+	}); err != nil {
+		t.Fatalf("UpdateHeartbeat: %v", err)
+	}
+
+	if err := ns.Delete(node.Id); err != nil {
+		t.Fatalf("Delete: %v", err)
+	}
+
+	prefix := nodeMetricKey(node.Id, "")
+	var left []string
+	nodeMetrics.mu.Lock()
+	for key := range nodeMetrics.series {
+		if strings.HasPrefix(key, prefix) {
+			left = append(left, key)
+		}
+	}
+	nodeMetrics.mu.Unlock()
+	if len(left) != 0 {
+		t.Fatalf("metric series left after deleting the node: %v", left)
+	}
+}